OpenAI proposed giving the U.S. government a 5% equity stake in the company, worth roughly $42.6 billion at the company’s current $852 billion valuation, according to the Financial Times on July 2, a report confirmed by Bloomberg (Bloomberg) and CNBC (CNBC). CEO Sam Altman raised the idea with President Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent; the envisioned structure would extend to other leading U.S. AI labs, with Anthropic, Google, and Meta each contributing comparable stakes to a sovereign AI wealth fund (The Next Web). The discussions are described as conceptual and early-stage, with any formal arrangement likely requiring an act of Congress. The proposal surfaces days after the Trump administration restricted access to GPT-5.6 Sol, Terra, and Luna to roughly 20 government-vetted partner organizations, leaving most developers without access to OpenAI’s most capable preview models (Tom’s Hardware).
Meta’s Chief AI Officer Alexandr Wang told company employees on July 2 that the company’s in-training model, code-named Watermelon, has matched OpenAI’s GPT-5.5 on the benchmarks Meta tracks internally, according to reporting from Business Insider (American Bazaar Online, Benzinga). Wang said Watermelon uses an order of magnitude more compute than Avocado - the internal codename for Meta’s previously released Muse Spark flagship - but did not name the specific benchmarks or release external verification data (Benzinga). Meta declined to comment and OpenAI did not respond to the report. If the internal claim holds on independent evaluation, it would mark a substantial narrowing of Meta’s gap to the frontier, though Watermelon remains in training with no confirmed release date, and GPT-5.5 is already one generation behind OpenAI’s current preview models (AI Weekly).
The GPT-5.6 access restriction stems from a June 2 executive order requiring developers of covered frontier models to provide the federal government up to 30 days of pre-release access before any broader partner rollout (The White House). GPT-5.6 is the first model to be publicly gated under this framework; OpenAI said in a published statement that the arrangement “should not become the long-term default,” though the company complied with the request (TechCrunch). A bipartisan discussion draft called the Great American AI Act, released June 4 by Reps. Jay Obernolte and Lori Trahan, proposes codifying similar frontier model review requirements in statute while preempting state laws that specifically regulate model development for three years (Rep. Jay Obernolte, Roll Call). That draft remains at the discussion stage.